Examining the Impact of Sample Size in the Analysis of Bicycle Sharing Systems

Given the growing installation of bicycle-sharing systems (BSS) across the world, there is a gradual increase in research on BSS over the past few years. Research efforts examining BSS employed a wide range of sample size depending on the temporal or spatial aggregation. The main objective of this paper is to investigate the impact of sample size on BSS analysis using data from New York City’s BSS (CitiBike). This paper proposes a systematic evaluation of the impact of sample size on model estimates, inference measures and predictive performance. We evaluate two major dimensions of BSS data: 1) system usage – impact of contributing factors on hourly arrival and departure rates at station level, 2) user destination choice – impact of factors on users’ preference of destination station choice. To examine the system usage, we employ the linear mixed model methodology while the user destination choice is studied using the Multinomial Logit Model (MNL). The model estimation exercises for system demand and destination choice are conducted on several samples of data. The performance of these sample models in terms of parameters, inference statistics and predictions relative to a base sample data is observed. The results would help the analysts to make decisions on sample size for accurately examining BSS usage. The analyses show that the impact of sample size on parameters estimated is stronger than that of the impact on prediction performance.